void SSearchTest::sharpSTest()
{
UErrorCode status = U_ZERO_ERROR;
UCollator *coll = NULL;
UnicodeString lp = "fuss";
UnicodeString sp = "fu\\u00DF";
UnicodeString targets[] = {"fu\\u00DF", "fu\\u00DFball", "1fu\\u00DFball", "12fu\\u00DFball", "123fu\\u00DFball", "1234fu\\u00DFball",
"ffu\\u00DF", "fufu\\u00DF", "fusfu\\u00DF",
"fuss", "ffuss", "fufuss", "fusfuss", "1fuss", "12fuss", "123fuss", "1234fuss", "fu\\u00DF", "1fu\\u00DF", "12fu\\u00DF", "123fu\\u00DF", "1234fu\\u00DF"};
int32_t start = -1, end = -1;
coll = ucol_openFromShortString("LEN_S1", FALSE, NULL, &status);
TEST_ASSERT_SUCCESS(status);
UnicodeString lpUnescaped = lp.unescape();
UnicodeString spUnescaped = sp.unescape();
LocalUStringSearchPointer ussLong(usearch_openFromCollator(lpUnescaped.getBuffer(), lpUnescaped.length(),
lpUnescaped.getBuffer(), lpUnescaped.length(), // actual test data will be set later
coll,
NULL, // the break iterator
&status));
LocalUStringSearchPointer ussShort(usearch_openFromCollator(spUnescaped.getBuffer(), spUnescaped.length(),
spUnescaped.getBuffer(), spUnescaped.length(), // actual test data will be set later
coll,
NULL, // the break iterator
&status));
TEST_ASSERT_SUCCESS(status);
for (uint32_t t = 0; t < (sizeof(targets)/sizeof(targets[0])); t += 1) {
UBool bFound;
UnicodeString target = targets[t].unescape();
start = end = -1;
usearch_setText(ussLong.getAlias(), target.getBuffer(), target.length(), &status);
bFound = usearch_search(ussLong.getAlias(), 0, &start, &end, &status);
TEST_ASSERT_SUCCESS(status);
if (bFound) {
logln("Test %d: found long pattern at [%d, %d].", t, start, end);
} else {
dataerrln("Test %d: did not find long pattern.", t);
}
usearch_setText(ussShort.getAlias(), target.getBuffer(), target.length(), &status);
bFound = usearch_search(ussShort.getAlias(), 0, &start, &end, &status);
TEST_ASSERT_SUCCESS(status);
if (bFound) {
logln("Test %d: found long pattern at [%d, %d].", t, start, end);
} else {
dataerrln("Test %d: did not find long pattern.", t);
}
}
ucol_close(coll);
}
void SSearchTest::searchTest()
{
#if !UCONFIG_NO_REGULAR_EXPRESSIONS && !UCONFIG_NO_FILE_IO
UErrorCode status = U_ZERO_ERROR;
char path[PATH_BUFFER_SIZE];
const char *testFilePath = getPath(path, "ssearch.xml");
if (testFilePath == NULL) {
return; /* Couldn't get path: error message already output. */
}
LocalPointer<UXMLParser> parser(UXMLParser::createParser(status));
TEST_ASSERT_SUCCESS(status);
LocalPointer<UXMLElement> root(parser->parseFile(testFilePath, status));
TEST_ASSERT_SUCCESS(status);
if (U_FAILURE(status)) {
return;
}
const UnicodeString *debugTestCase = root->getAttribute("debug");
if (debugTestCase != NULL) {
// setenv("USEARCH_DEBUG", "1", 1);
}
const UXMLElement *testCase;
int32_t tc = 0;
while((testCase = root->nextChildElement(tc)) != NULL) {
if (testCase->getTagName().compare("test-case") != 0) {
errln("ssearch, unrecognized XML Element in test file");
continue;
}
const UnicodeString *id = testCase->getAttribute("id");
*testId = 0;
if (id != NULL) {
id->extract(0, id->length(), testId, sizeof(testId), US_INV);
}
// If debugging test case has been specified and this is not it, skip to next.
if (id!=NULL && debugTestCase!=NULL && *id != *debugTestCase) {
continue;
}
//
// Get the requested collation strength.
// Default is tertiary if the XML attribute is missing from the test case.
//
const UnicodeString *strength = testCase->getAttribute("strength");
UColAttributeValue collatorStrength = UCOL_PRIMARY;
if (strength==NULL) { collatorStrength = UCOL_TERTIARY;}
else if (*strength=="PRIMARY") { collatorStrength = UCOL_PRIMARY;}
else if (*strength=="SECONDARY") { collatorStrength = UCOL_SECONDARY;}
else if (*strength=="TERTIARY") { collatorStrength = UCOL_TERTIARY;}
else if (*strength=="QUATERNARY") { collatorStrength = UCOL_QUATERNARY;}
else if (*strength=="IDENTICAL") { collatorStrength = UCOL_IDENTICAL;}
else {
// Bogus value supplied for strength. Shouldn't happen, even from
// typos, if the XML source has been validated.
// This assert is a little deceiving in that strength can be
// any of the allowed values, not just TERTIARY, but it will
// do the job of getting the error output.
TEST_ASSERT(*strength=="TERTIARY")
}
//
// Get the collator normalization flag. Default is UCOL_OFF.
//
UColAttributeValue normalize = UCOL_OFF;
const UnicodeString *norm = testCase->getAttribute("norm");
TEST_ASSERT (norm==NULL || *norm=="ON" || *norm=="OFF");
if (norm!=NULL && *norm=="ON") {
normalize = UCOL_ON;
}
//
// Get the alternate_handling flag. Default is UCOL_NON_IGNORABLE.
//
UColAttributeValue alternateHandling = UCOL_NON_IGNORABLE;
const UnicodeString *alt = testCase->getAttribute("alternate_handling");
TEST_ASSERT (alt == NULL || *alt == "SHIFTED" || *alt == "NON_IGNORABLE");
if (alt != NULL && *alt == "SHIFTED") {
alternateHandling = UCOL_SHIFTED;
}
const UnicodeString defLocale("en");
char clocale[100];
const UnicodeString *locale = testCase->getAttribute("locale");
if (locale == NULL || locale->length()==0) {
locale = &defLocale;
};
locale->extract(0, locale->length(), clocale, sizeof(clocale), NULL);
UnicodeString text;
UnicodeString target;
UnicodeString pattern;
int32_t expectedMatchStart = -1;
int32_t expectedMatchLimit = -1;
const UXMLElement *n;
int32_t nodeCount = 0;
n = testCase->getChildElement("pattern");
TEST_ASSERT(n != NULL);
if (n==NULL) {
continue;
}
text = n->getText(FALSE);
text = text.unescape();
pattern.append(text);
nodeCount++;
n = testCase->getChildElement("pre");
if (n!=NULL) {
text = n->getText(FALSE);
text = text.unescape();
target.append(text);
nodeCount++;
}
n = testCase->getChildElement("m");
if (n!=NULL) {
expectedMatchStart = target.length();
text = n->getText(FALSE);
text = text.unescape();
target.append(text);
expectedMatchLimit = target.length();
nodeCount++;
}
n = testCase->getChildElement("post");
if (n!=NULL) {
text = n->getText(FALSE);
text = text.unescape();
target.append(text);
nodeCount++;
}
// Check that there weren't extra things in the XML
TEST_ASSERT(nodeCount == testCase->countChildren());
// Open a collator and StringSearch based on the parameters
// obtained from the XML.
//
status = U_ZERO_ERROR;
LocalUCollatorPointer collator(ucol_open(clocale, &status));
ucol_setStrength(collator.getAlias(), collatorStrength);
ucol_setAttribute(collator.getAlias(), UCOL_NORMALIZATION_MODE, normalize, &status);
ucol_setAttribute(collator.getAlias(), UCOL_ALTERNATE_HANDLING, alternateHandling, &status);
LocalUStringSearchPointer uss(usearch_openFromCollator(pattern.getBuffer(), pattern.length(),
target.getBuffer(), target.length(),
collator.getAlias(),
NULL, // the break iterator
&status));
TEST_ASSERT_SUCCESS(status);
if (U_FAILURE(status)) {
continue;
}
int32_t foundStart = 0;
int32_t foundLimit = 0;
UBool foundMatch;
//
// Do the search, check the match result against the expected results.
//
foundMatch= usearch_search(uss.getAlias(), 0, &foundStart, &foundLimit, &status);
TEST_ASSERT_SUCCESS(status);
if ((foundMatch && expectedMatchStart<0) ||
(foundStart != expectedMatchStart) ||
(foundLimit != expectedMatchLimit)) {
TEST_ASSERT(FALSE); // ouput generic error position
infoln("Found, expected match start = %d, %d \n"
"Found, expected match limit = %d, %d",
foundStart, expectedMatchStart, foundLimit, expectedMatchLimit);
}
// In case there are other matches...
// (should we only do this if the test case passed?)
while (foundMatch) {
expectedMatchStart = foundStart;
expectedMatchLimit = foundLimit;
foundMatch = usearch_search(uss.getAlias(), foundLimit, &foundStart, &foundLimit, &status);
}
uss.adoptInstead(usearch_openFromCollator(pattern.getBuffer(), pattern.length(),
target.getBuffer(), target.length(),
collator.getAlias(),
NULL,
&status));
//
// Do the backwards search, check the match result against the expected results.
//
foundMatch= usearch_searchBackwards(uss.getAlias(), target.length(), &foundStart, &foundLimit, &status);
TEST_ASSERT_SUCCESS(status);
if ((foundMatch && expectedMatchStart<0) ||
(foundStart != expectedMatchStart) ||
(foundLimit != expectedMatchLimit)) {
TEST_ASSERT(FALSE); // ouput generic error position
infoln("Found, expected backwards match start = %d, %d \n"
"Found, expected backwards match limit = %d, %d",
foundStart, expectedMatchStart, foundLimit, expectedMatchLimit);
}
}
#endif
}
void
JamoTest::TestRealText() {
// Test text taken from the Unicode web site
static const char* const WHAT_IS_UNICODE[] = {
"\\uc720\\ub2c8\\ucf54\\ub4dc\\uc5d0", "\\ub300\\ud574", "?",
"\\uc5b4\\ub5a4", "\\ud50c\\ub7ab\\ud3fc,", "\\uc5b4\\ub5a4",
"\\ud504\\ub85c\\uadf8\\ub7a8,", "\\uc5b4\\ub5a4", "\\uc5b8\\uc5b4\\uc5d0\\ub3c4",
"\\uc0c1\\uad00\\uc5c6\\uc774", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294", "\\ubaa8\\ub4e0",
"\\ubb38\\uc790\\uc5d0", "\\ub300\\ud574", "\\uace0\\uc720", "\\ubc88\\ud638\\ub97c",
"\\uc81c\\uacf5\\ud569\\ub2c8\\ub2e4.",
"\\uae30\\ubcf8\\uc801\\uc73c\\ub85c", "\\ucef4\\ud4e8\\ud130\\ub294",
"\\uc22b\\uc790\\ub9cc", "\\ucc98\\ub9ac\\ud569\\ub2c8\\ub2e4.", "\\uae00\\uc790\\ub098",
"\\ub2e4\\ub978", "\\ubb38\\uc790\\uc5d0\\ub3c4", "\\uc22b\\uc790\\ub97c",
"\\uc9c0\\uc815\\ud558\\uc5ec",
"\\uc800\\uc7a5\\ud569\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\uac00",
"\\uac1c\\ubc1c\\ub418\\uae30", "\\uc804\\uc5d0\\ub294", "\\uc774\\ub7ec\\ud55c",
"\\uc22b\\uc790\\ub97c", "\\uc9c0\\uc815\\ud558\\uae30", "\\uc704\\ud574", "\\uc218\\ubc31",
"\\uac00\\uc9c0\\uc758", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654",
"\\uc2dc\\uc2a4\\ud15c\\uc744",
"\\uc0ac\\uc6a9\\ud588\\uc2b5\\ub2c8\\ub2e4.", "\\ub2e8\\uc77c", "\\uae30\\ud638\\ud654",
"\\ubc29\\ubc95\\uc73c\\ub85c\\ub294", "\\ubaa8\\ub4e0", "\\ubb38\\uc790\\ub97c",
"\\ud3ec\\ud568\\ud560", "\\uc218", "\\uc5c6\\uc5c8\\uc2b5\\ub2c8\\ub2e4.", "\\uc608\\ub97c",
"\\ub4e4\\uc5b4", "\\uc720\\ub7fd", "\\uc5f0\\ud569\\uc5d0\\uc11c\\ub9cc",
"\\ubcf4\\ub354\\ub77c\\ub3c4", "\\ubaa8\\ub4e0", "\\uac01", "\\ub098\\ub77c\\ubcc4",
"\\uc5b8\\uc5b4\\ub97c", "\\ucc98\\ub9ac\\ud558\\ub824\\uba74", "\\uc5ec\\ub7ec",
"\\uac1c\\uc758", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774",
"\\ud544\\uc694\\ud569\\ub2c8\\ub2e4.", "\\uc601\\uc5b4\\uc640", "\\uac19\\uc740",
"\\ub2e8\\uc77c", "\\uc5b8\\uc5b4\\uc758", "\\uacbd\\uc6b0\\ub3c4",
"\\uacf5\\ud1b5\\uc801\\uc73c\\ub85c", "\\uc0ac\\uc6a9\\ub418\\ub294", "\\ubaa8\\ub4e0",
"\\uae00\\uc790,", "\\ubb38\\uc7a5", "\\ubd80\\ud638", "\\ubc0f",
"\\ud14c\\ud06c\\ub2c8\\uceec", "\\uae30\\ud638\\uc5d0", "\\ub9de\\ub294", "\\ub2e8\\uc77c",
"\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc744", "\\uac16\\uace0", "\\uc788\\uc9c0",
"\\ubabb\\ud558\\uc600\\uc2b5\\ub2c8\\ub2e4.",
"\\uc774\\ub7ec\\ud55c", "\\uae30\\ud638\\ud654", "\\uc2dc\\uc2a4\\ud15c\\uc740",
"\\ub610\\ud55c", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\uc2dc\\uc2a4\\ud15c\\uacfc",
"\\ucda9\\ub3cc\\ud569\\ub2c8\\ub2e4.", "\\uc989", "\\ub450", "\\uac00\\uc9c0",
"\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774", "\\ub450", "\\uac1c\\uc758", "\\ub2e4\\ub978",
"\\ubb38\\uc790\\uc5d0", "\\ub300\\ud574", "\\uac19\\uc740", "\\ubc88\\ud638\\ub97c",
"\\uc0ac\\uc6a9\\ud558\\uac70\\ub098", "\\uac19\\uc740", "\\ubb38\\uc790\\uc5d0",
"\\ub300\\ud574", "\\ub2e4\\ub978", "\\ubc88\\ud638\\ub97c", "\\uc0ac\\uc6a9\\ud560", "\\uc218",
"\\uc788\\uc2b5\\ub2c8\\ub2e4.", "\\uc8fc\\uc5b4\\uc9c4", "\\ubaa8\\ub4e0",
"\\ucef4\\ud4e8\\ud130(\\ud2b9\\ud788", "\\uc11c\\ubc84)\\ub294", "\\uc11c\\ub85c",
"\\ub2e4\\ub978", "\\uc5ec\\ub7ec", "\\uac00\\uc9c0", "\\uae30\\ud638\\ud654",
"\\ubc29\\ubc95\\uc744", "\\uc9c0\\uc6d0\\ud574\\uc57c",
"\\ud569\\ub2c8\\ub2e4.", "\\uadf8\\ub7ec\\ub098,", "\\ub370\\uc774\\ud130\\ub97c",
"\\uc11c\\ub85c", "\\ub2e4\\ub978", "\\uae30\\ud638\\ud654", "\\ubc29\\ubc95\\uc774\\ub098",
"\\ud50c\\ub7ab\\ud3fc", "\\uac04\\uc5d0", "\\uc804\\ub2ec\\ud560", "\\ub54c\\ub9c8\\ub2e4",
"\\uadf8", "\\ub370\\uc774\\ud130\\ub294", "\\ud56d\\uc0c1", "\\uc190\\uc0c1\\uc758",
"\\uc704\\ud5d8\\uc744", "\\uacaa\\uac8c", "\\ub429\\ub2c8\\ub2e4.",
"\\uc720\\ub2c8\\ucf54\\ub4dc\\ub85c", "\\ubaa8\\ub4e0", "\\uac83\\uc744",
"\\ud574\\uacb0\\ud560", "\\uc218", "\\uc788\\uc2b5\\ub2c8\\ub2e4!",
"\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294", "\\uc0ac\\uc6a9", "\\uc911\\uc778",
"\\ud50c\\ub7ab\\ud3fc,", "\\ud504\\ub85c\\uadf8\\ub7a8,", "\\uc5b8\\uc5b4\\uc5d0",
"\\uad00\\uacc4\\uc5c6\\uc774", "\\ubb38\\uc790\\ub9c8\\ub2e4", "\\uace0\\uc720\\ud55c",
"\\uc22b\\uc790\\ub97c",
"\\uc81c\\uacf5\\ud569\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
"\\ud45c\\uc900\\uc740", // "Apple, HP, IBM, JustSystem, Microsoft, Oracle, SAP, "
// "Sun, Sybase, Unisys "
"\\ubc0f", "\\uae30\\ud0c0", "\\uc5ec\\ub7ec",
"\\ud68c\\uc0ac\\uc640", "\\uac19\\uc740", "\\uc5c5\\uacc4",
"\\uc120\\ub450\\uc8fc\\uc790\\uc5d0", "\\uc758\\ud574",
"\\ucc44\\ud0dd\\ub418\\uc5c8\\uc2b5\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub294",
// "XML, Java, ECMAScript(JavaScript), LDAP, CORBA 3.0, WML "
"\\ub4f1\\uacfc",
"\\uac19\\uc774", "\\ud604\\uc7ac", "\\ub110\\ub9ac", "\\uc0ac\\uc6a9\\ub418\\ub294",
"\\ud45c\\uc900\\uc5d0\\uc11c", "\\ud544\\uc694\\ud558\\uba70", "\\uc774\\ub294", // "ISO/IEC",
"10646\\uc744", "\\uad6c\\ud604\\ud558\\ub294", "\\uacf5\\uc2dd\\uc801\\uc778",
"\\ubc29\\ubc95\\uc785\\ub2c8\\ub2e4.", "\\uc774\\ub294", "\\ub9ce\\uc740", "\\uc6b4\\uc601",
"\\uccb4\\uc81c,", "\\uc694\\uc998", "\\uc0ac\\uc6a9\\ub418\\ub294", "\\ubaa8\\ub4e0",
"\\ube0c\\ub77c\\uc6b0\\uc800", "\\ubc0f", "\\uae30\\ud0c0", "\\ub9ce\\uc740",
"\\uc81c\\ud488\\uc5d0\\uc11c",
"\\uc9c0\\uc6d0\\ub429\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
"\\ud45c\\uc900\\uc758", "\\ubd80\\uc0c1\\uacfc", "\\uc774\\ub97c",
"\\uc9c0\\uc6d0\\ud558\\ub294", "\\ub3c4\\uad6c\\uc758", "\\uac00\\uc6a9\\uc131\\uc740",
"\\ucd5c\\uadfc", "\\uc804", "\\uc138\\uacc4\\uc5d0", "\\ubd88\\uace0", "\\uc788\\ub294",
"\\uae30\\uc220", "\\uacbd\\ud5a5\\uc5d0\\uc11c", "\\uac00\\uc7a5", "\\uc911\\uc694\\ud55c",
"\\ubd80\\ubd84\\uc744", "\\ucc28\\uc9c0\\ud558\\uace0", "\\uc788\\uc2b5\\ub2c8\\ub2e4.",
"\\uc720\\ub2c8\\ucf54\\ub4dc\\ub97c",
// Replaced a hyphen with a space to make the test case work with CLDR1.5
//"\\ud074\\ub77c\\uc774\\uc5b8\\ud2b8-\\uc11c\\ubc84", "\\ub610\\ub294",
"\\ud074\\ub77c\\uc774\\uc5b8\\ud2b8 \\uc11c\\ubc84", "\\ub610\\ub294",
// Replaced a hyphen with a space.
//"\\ub2e4\\uc911-\\uc5f0\\uacb0", "\\uc751\\uc6a9", "\\ud504\\ub85c\\uadf8\\ub7a8\\uacfc",
"\\ub2e4\\uc911 \\uc5f0\\uacb0", "\\uc751\\uc6a9", "\\ud504\\ub85c\\uadf8\\ub7a8\\uacfc",
"\\uc6f9", "\\uc0ac\\uc774\\ud2b8\\uc5d0", "\\ud1b5\\ud569\\ud558\\uba74",
"\\ub808\\uac70\\uc2dc", "\\ubb38\\uc790", "\\uc138\\ud2b8", "\\uc0ac\\uc6a9\\uc5d0",
"\\uc788\\uc5b4\\uc11c", "\\uc0c1\\ub2f9\\ud55c", "\\ube44\\uc6a9", "\\uc808\\uac10",
"\\ud6a8\\uacfc\\uac00",
"\\ub098\\ud0c0\\ub0a9\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc\\ub97c",
"\\ud1b5\\ud574", "\\ub9ac\\uc5d4\\uc9c0\\ub2c8\\uc5b4\\ub9c1", "\\uc5c6\\uc774",
"\\ub2e4\\uc911", "\\ud50c\\ub7ab\\ud3fc,", "\\uc5b8\\uc5b4", "\\ubc0f", "\\uad6d\\uac00",
"\\uac04\\uc5d0", "\\ub2e8\\uc77c", "\\uc18c\\ud504\\ud2b8\\uc6e8\\uc5b4",
"\\ud50c\\ub7ab\\ud3fc", "\\ub610\\ub294", "\\ub2e8\\uc77c", "\\uc6f9",
"\\uc0ac\\uc774\\ud2b8\\ub97c", "\\ubaa9\\ud45c\\ub85c", "\\uc0bc\\uc744", "\\uc218",
"\\uc788\\uc2b5\\ub2c8\\ub2e4.", "\\uc774\\ub97c", "\\uc0ac\\uc6a9\\ud558\\uba74",
"\\ub370\\uc774\\ud130\\ub97c", "\\uc190\\uc0c1", "\\uc5c6\\uc774", "\\uc5ec\\ub7ec",
"\\uc2dc\\uc2a4\\ud15c\\uc744", "\\ud1b5\\ud574", "\\uc804\\uc1a1\\ud560", "\\uc218",
"\\uc788\\uc2b5\\ub2c8\\ub2e4.",
"\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc5d0", "\\ub300\\ud574",
"\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc740",
"\\ube44\\uc601\\ub9ac", "\\uc870\\uc9c1\\uc73c\\ub85c\\uc11c", "\\ud604\\ub300",
"\\uc18c\\ud504\\ud2b8\\uc6e8\\uc5b4", "\\uc81c\\ud488\\uacfc",
"\\ud45c\\uc900\\uc5d0\\uc11c", "\\ud14d\\uc2a4\\ud2b8\\uc758", "\\ud45c\\ud604\\uc744",
"\\uc9c0\\uc815\\ud558\\ub294", "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ud45c\\uc900\\uc758",
"\\uc0ac\\uc6a9\\uc744", "\\uac1c\\ubc1c\\ud558\\uace0", "\\ud655\\uc7a5\\ud558\\uba70",
"\\uc7a5\\ub824\\ud558\\uae30", "\\uc704\\ud574",
"\\uc138\\uc6cc\\uc84c\\uc2b5\\ub2c8\\ub2e4.", "\\ucf58\\uc18c\\uc2dc\\uc5c4",
"\\uba64\\ubc84\\uc27d\\uc740", "\\ucef4\\ud4e8\\ud130\\uc640", "\\uc815\\ubcf4",
"\\ucc98\\ub9ac", "\\uc0b0\\uc5c5\\uc5d0", "\\uc885\\uc0ac\\ud558\\uace0", "\\uc788\\ub294",
"\\uad11\\ubc94\\uc704\\ud55c", "\\ud68c\\uc0ac", "\\ubc0f", "\\uc870\\uc9c1\\uc758",
"\\ubc94\\uc704\\ub97c",
"\\ub098\\ud0c0\\ub0c5\\ub2c8\\ub2e4.", "\\ucf58\\uc18c\\uc2dc\\uc5c4\\uc758",
"\\uc7ac\\uc815\\uc740", "\\uc804\\uc801\\uc73c\\ub85c", "\\ud68c\\ube44\\uc5d0",
"\\uc758\\ud574", "\\ucda9\\ub2f9\\ub429\\ub2c8\\ub2e4.", "\\uc720\\ub2c8\\ucf54\\ub4dc",
"\\ucee8\\uc18c\\uc2dc\\uc5c4\\uc5d0\\uc11c\\uc758", "\\uba64\\ubc84\\uc27d\\uc740",
"\\uc804", "\\uc138\\uacc4", "\\uc5b4\\ub290", "\\uacf3\\uc5d0\\uc11c\\ub098",
"\\uc720\\ub2c8\\ucf54\\ub4dc", "\\ud45c\\uc900\\uc744", "\\uc9c0\\uc6d0\\ud558\\uace0",
"\\uadf8", "\\ud655\\uc7a5\\uacfc", "\\uad6c\\ud604\\uc744",
"\\uc9c0\\uc6d0\\ud558\\uace0\\uc790\\ud558\\ub294", "\\uc870\\uc9c1\\uacfc",
"\\uac1c\\uc778\\uc5d0\\uac8c", "\\uac1c\\ubc29\\ub418\\uc5b4",
"\\uc788\\uc2b5\\ub2c8\\ub2e4.",
"\\ub354", "\\uc790\\uc138\\ud55c", "\\ub0b4\\uc6a9\\uc740", "\\uc6a9\\uc5b4\\uc9d1,",
"\\uc608\\uc81c", "\\uc720\\ub2c8\\ucf54\\ub4dc", "\\uc0ac\\uc6a9", "\\uac00\\ub2a5",
"\\uc81c\\ud488,", "\\uae30\\uc220", "\\uc815\\ubcf4", "\\ubc0f", "\\uae30\\ud0c0",
"\\uc720\\uc6a9\\ud55c", "\\uc815\\ubcf4\\ub97c",
"\\ucc38\\uc870\\ud558\\uc2ed\\uc2dc\\uc624."
};
enum { WHAT_IS_UNICODE_length = sizeof(WHAT_IS_UNICODE) / sizeof(WHAT_IS_UNICODE[0]) };
UParseError parseError;
UErrorCode status = U_ZERO_ERROR;
Transliterator* latinJamo = Transliterator::createInstance("Latin-Jamo", UTRANS_FORWARD, parseError, status);
Transliterator* jamoHangul = Transliterator::createInstance("NFC(NFD)", UTRANS_FORWARD, parseError, status);
if (latinJamo == 0 || jamoHangul == 0 || U_FAILURE(status)) {
delete latinJamo;
delete jamoHangul;
dataerrln("FAIL: createInstance returned NULL - %s", u_errorName(status));
return;
}
Transliterator* jamoLatin = latinJamo->createInverse(status);
Transliterator* hangulJamo = jamoHangul->createInverse(status);
if (jamoLatin == 0 || hangulJamo == 0) {
errln("FAIL: createInverse returned NULL");
delete latinJamo;
delete jamoLatin;
delete jamoHangul;
delete hangulJamo;
return;
}
Transliterator* tarray[4] =
{ hangulJamo, jamoLatin, latinJamo, jamoHangul };
CompoundTransliterator rt(tarray, 4);
UnicodeString buf;
int32_t total = 0;
int32_t errors = 0;
int32_t i;
for (i=0; i < WHAT_IS_UNICODE_length; ++i) {
++total;
UnicodeString hangul = UnicodeString(WHAT_IS_UNICODE[i], -1, US_INV);
hangul = hangul.unescape(); // Parse backslash-u escapes
UnicodeString hangulX = hangul;
rt.transliterate(hangulX);
if (hangul != hangulX) {
++errors;
UnicodeString jamo = hangul; hangulJamo->transliterate(jamo);
UnicodeString latin = jamo; jamoLatin->transliterate(latin);
UnicodeString jamo2 = latin; latinJamo->transliterate(jamo2);
UnicodeString hangul2 = jamo2; jamoHangul->transliterate(hangul2);
buf.remove(0);
buf.append("FAIL: ");
if (hangul2 != hangulX) {
buf.append((UnicodeString)"(Weird: " + hangulX + " != " + hangul2 + ")");
}
// The Hangul-Jamo conversion is not usually the
// bug here, so we hide it from display.
// Uncomment lines to see the Hangul.
buf.append(//hangul + " => " +
jamoToName(jamo) + " => " +
latin + " => " + jamoToName(jamo2)
//+ " => " + hangul2
);
errln(prettify(buf));
}
}
if (errors != 0) {
errln((UnicodeString)"Test word failures: " + errors + " out of " + total);
} else {
logln((UnicodeString)"All " + total + " test words passed");
}
delete latinJamo;
delete jamoLatin;
delete jamoHangul;
delete hangulJamo;
}
